Post on 06-Aug-2020
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
1
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status
and Healthcare Quality Elissa R Weitzman,
1,2,4 ScD, MSc, Skyler Kelemen,
1BA, Kenneth D Mandl,
1,3,4,5 MD, MPH
1Children’s Hospital Informatics Program at the Harvard-MIT Division of Health
Sciences and Technology, Children’s Hospital Boston, Boston, MA, USA 2Division of Adolescent Medicine, Children’s Hospital Boston, Boston, MA, USA
3Division of Emergency Medicine, Children’s Hospital Boston, Boston, MA, USA
4Department of Pediatrics, Harvard Medical School, Boston, MA, USA
5Manton Center for Orphan Disease Research, Children’s Hospital Boston, MA
Abstract
Objective: Test a novel health monitoring approach by engaging an international online
diabetes social network (SN) in consented health surveillance.
Methods: Collection of structured self-reports about preventive and self-care practices and
health status using a software application (“app”) that supports SN-mediated health research.
Comparison of SN measures by diabetes type; and, SN with Behavioral Risk Factor
Surveillance System (BRFSS) data, for US-residing insulin dependent respondents, using
logistic regression.
Results: Of 2,414 SN app users, 82% (n=1979) provided an A1c and 41% (n=996) completed a
care survey of which 931 have diabetes. Of these: 65% and 41% were immunized against
influenza and pneumonia respectively, 90% had their cholesterol checked, 82% and 66%, had
their eyes and feet checked, respectively. Type 1/LADA respondents were more likely than
Type 2/pre-diabetic respondents to report all five recommended practices (Adjusted OR (95%
CI) 2.2 (1.5, 3.2)). Past year self-care measures were: 58% self-monitored their blood glucose
(SMBG) ≥ 5 times daily, 37% saw their diabetes nutritionist, 56% saw a diabetes nurse
educator, 53% saw a doctor for their diabetes ≥ 4 times. Reports of health status did not differ
by diabetes type in the SN sample. The SN group was more likely than the BRFSS comparator
group to use all five preventive care practices (Adjusted OR (95% CI) 1.8 (1.4, 2.1) and SMBG
≥ 5 times daily (Adjusted OR (95% CI) 10.1 (6.8, 14.9).
Conclusions: Rapid assessment of diabetes care practices using a novel, SN-mediated
approach can extend the capability of standard health surveillance systems.
Keywords: diabetes, healthcare quality, social networks, surveillance, social networking,
chronic illness
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
2
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Introduction
Traditional health surveillance programs, for example the Behavioral Risk Factor Surveillance
System (BRFSS) and the National Health and Nutrition Examination Survey (NHANES),
depend on increasingly unreliable channels for communication such as landline-based telephone
surveys and household interviews. These approaches have enriched our understanding of
population patterns of disease and illuminated the increasing morbidity and mortality burdens
imposed by chronic illnesses (1, 2). However, standard approaches face challenges to obtain and
retain participants, accommodate rapidly shifting patterns of disease, and balance breadth with
depth of data collected in a fashion that supports assessment of disease prevalence and also
healthcare and self-care practices of important affected subgroups (3).
Recently, we reported about our efforts to engage participants in an international online diabetes
social network (SN) in consented public health monitoring of their disease (4). Using a software
application (“app”) we launched into their Facebook-like community, SN members could report
about and share their diabetes health data and participate in research as part of a distributed
public health research cohort. The project falls within the rubric of new ‘citizen science’ efforts
(5) to engage populations—including online social networks—in advancing public health
through contributing data and observational energies to research (6, 7). This SN- mediated
approach to health monitoring may address some of the challenges facing standard surveillance
systems by engaging a population of interest in bidirectional communication about important and
often overlooked aspects of their disease vital to targeted interventions.
For this report, we investigated diabetes health status and adherence with diabetes-specific
recommended preventive and self-care practices among SN members using this novel app and
approach. A primary focus of the investigation was to characterize care patterns in the sample
overall and by diabetes type, an important stratification variable for understanding diabetes—a
heterogeneous disorder originating in different biologic and sociologic processes (8-10). Type
data are not collected in BRFSS (11) nor NHANES (12), limiting the utility of these systems for
informing targeted response. A secondary focus of the investigation was to compare care metrics
reported by the SN sample to those reported by respondents from a standard health surveillance
system to ascertain the extent to which patterns resemble each other.
Methods
We used the TuAnalyze app (4) to survey participating members of the TuDiabetes community
about their preventive and self-care practices. Members report and share their diabetes data and
obtain both contextualized views of personal measures of glcyemic control (A1c%) (Figure 1),
and summary reports about the health status and care patterns of all application users (Figure 2).
The application is available in English and Spanish language versions. Its use is voluntary. Study
activities were reviewed and approved by the Children’s Hospital Boston Institutional Review
Board. Details about the application design, technology platform, operations and early
adoption/use patterns are published elsewhere (4).
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
3
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Figure 1. Contextualized map view
of A1c% data visible in the
TuAnalyze application.
Users of the TuAnalyze application
who self-report measures of glycemic
control (A1c%) using the application
are able to view near real-time
summary data of their A1c% charted
against the frequency distribution of
other members’ shared A1c%
measures, by geographic area (state,
province or country). Map views of
contextualized A1c% measures are
delivered on a geographic information
system display in which only a given
user can see her personal data arrayed
against summary data.
Figure 2. Illustration of a
TuAnalyze blog post and feedback
report in which aggregate
TuAnalyze data were shared with
the TuDiabetes community and
TuAnalyze users.
In the TuAnalyze application,
members of the TuDiabetes.org host
online social network as well as
users of the TuAnalyze app, can read
summary reports generated from
data entered into the application by
the community of users and shared
as a means of communicating
“results” and sparking engagement
with the technology. Comments and
questions posted by users are
reviewed and incorporated into the
study agenda as part of the
participatory research approach.
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
4
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
TuAnalyze Sample Eligibility
Eligibility criteria for TuAnalyze and the larger TuDiabetes social network include being age 18
or older (younger users may join with a parent/guardian), affected by diabetes, ability to read and
write English and/or Spanish, and having Internet access. Persons who do not have diabetes and
are using the application as a proxy for another person, such as family members (N=65), were
not included in these analyses.
Response Rate
During the study period, 996 out of 2,414 TuAnalyze users (response rate 41.2%) took a survey
about their diabetes and associated care patterns, a response rate considered high for web surveys
(13). The final sample consisted of 931 users who had completed the survey and have some form
of diabetes. 105 users completed the survey but did not enter an A1c measure into the
application; these were excluded from all analyses of A1c but otherwise included. Survey
respondents who had not entered an A1c value were less likely than those who had to be white,
to reside in the US, to have health insurance, and to use the English language version of the
application.
Measures
Analyses draw on two sets of data: self-reported data from TuAnalyze users including A1c%
values and surveys administered in the application environment; and self-reported health and
demographic data from a US national sample surveyed for the 2009 Behavioral Risk Factor
Surveillance System (BRFSS). The surveys administered in TuAnalyze were constructed from
questions adapted to the application and taken from national health surveys, including the
BRFSS (1, 12), and the National Health and Nutrition Examination Study (NHANES) (14).
TuAnalyze users indicated their diabetes type by selecting from a dropdown list in the survey
with options for self-identifying as having Type 1, Type 1.5 (LADA), Type 2, pre-diabetes and
gestational diabetes. We dichotomized diabetes type into two main groups, based on the
underlying disease mechanism, as Type 1 or LADA and Type 2 or pre-diabetes. No respondents
indicated gestational diabetes and this type is therefore not included.
Recommended diabetes preventive care practices were defined consistent with the 2010
American Diabetes Association (ADA) clinical practice recommendations, and included any
history of a pneumonia vaccination, and past year history of an influenza vaccination, dilated eye
exam, foot exam, and lipid profile (cholesterol check) (8). Self-care was assessed by reported
typical frequency of self-monitoring of blood glucose, dichotomized at the ADA minimum
recommended threshold of five or more times per day, and by describing reports of seeing a
diabetes nurse educator and nutritionist in the past year and having four or more visits with one’s
doctor in the past year.
TuAnalyze respondents were asked to rate their health on a standardized scale. Those who
reported “Excellent”, “Very Good” or “Good” health were classified as having better health,
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
5
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
compared to those who reported their health as “Fair” or “Poor”. Respondents who had ever
been diagnosed with one or more health problem besides depression from a list of chronic
conditions provided in the survey (arthritis, asthma, cancer, cardiovascular disease, depression,
stroke and “other”) were categorized as having a chronic illness comorbidity; those who reported
having been diagnosed with depression from this list were categorized as having experiencing
depression. Consistent with ADA standards (8), respondents with a most recent A1c value of
7% or higher were classified as “above target” compared to those with lower values.
For both the TuAnalyze and BRFSS cohorts, we classified as white respondents who identified
themselves solely as Caucasian and non-Hispanic. Age at the time of survey was determined
from respondents’ date of birth (for TuAnalyze) and reported age at time of survey (for BRFSS).
Country location for TuAnalyze users was gathered from data entered upon first engaging with
the application; all BRFSS respondents are based in the US.
Data analyses
Reported use patterns of diabetes-specific preventive and self-care practices among the
TuAnalyze sample were estimated in aggregate and across diabetes types. TuAnalyze patterns
were compared with those found in analyses of the 2009 US BRFSS (15), the most recent US
national health monitoring system for which comparable preventive and self-care measures are
available. BRFSS data were downloaded at no cost from the US Centers for Disease Control and
Prevention website (15). Descriptive statistics characterize the TuAnalyze sample and the Chi-
Squared test was used to compare demographics, care behaviors and health outcomes across the
two major diabetes type groups; this approach was also followed in comparing preventive and
self-care practices across the TuAnalyze and BRFSS samples. Comparative analyses of
TuAnalyze and BRFSS samples were undertaken for restricted samples that included
respondents with diabetes, who reside in the US and who take insulin—a proxy for disease type
given the absence of type data in the BRFSS. Cross-type and cross-sample comparisons are
reported for bivariate and multivariate logistic regression models, the latter control for effects of
sex, race (white/other), and age (continuous). Analyses were conducted in SAS version 9.2.
Results
Demographic characteristics of the TuAnalyze sample
As shown in Table 1, of 931 respondents, a majority is white, located in the US, and female.
Approximately two-thirds (62%) report having Type 1 diabetes and another 10% have Type 1.5
(LADA); 27% have type 2 diabetes and the remaining 1% report pre-diabetes. Age of the sample
ranged from 13 to 81, with an average and median of 43.
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
6
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Table 1. Characteristics of the TuAnalyze study sample, in aggregate and by type
N (%)
Total
sample
931 (100)
Type 1 &
LADA
664 (71.3)
Type 2/ Pre-
diabetes
267 (28.7)
OR for Type
1/LADA
English Language
site 797 (85.6) 594 (89.5) 203 (76) 2.7 (1.8, 3.9)***
US Location 664 (71.3) 505 (76.1) 167 (62.6) 1.6 (1.1, 2.2)**
White 745 (80) 564 (84.9) 181 (67.8) 2.7 (1.9, 3.4)***
Male 367 (39.4) 235 (35.4) 132 (49.4) .56 (.42, .75)***
Over 40 447 (54.1) 311 (46.8) 202 (75.7) .28 (.21, .39)***
*p<.05 **p<.01 ***p<.001
Use of recommended diabetes preventive and self care practices
Use of recommended preventive care practices was high to very high among respondents.
Upwards of four-fifths of the sample reported having an annual check for cholesterol and a check
for retinopathy in the past year. Upwards of two-thirds reported having an annual check of their
feet for circulatory problems and neuropathy, and an influenza vaccination, in the past year. In
contrast, two fifths reported having ever been immunized for pneumonia. Despite these
prevalence levels for individual care practices, less than one third of respondents reported
obtaining all five preventive care practices, a signifier of comprehensive preventive care.
Use of preventive care practices varied by diabetes type. Users reporting Type 1 or LADA
diabetes were more likely than their peers with other diabetes types to report they were ever
immunized for pneumonia and to report receiving an influenza vaccine, eye exam and foot exam
in the past year. There was no difference in reporting a cholesterol check in the past year by
diabetes type. Type 1 and LADA respondents had twice the odds of obtaining all five
recommended preventive care practices as Type 2 and pre-diabetes respondents, controlling for
age, sex and race (adjusted OR 2.2, 95% CI 1.5, 3.2, p <.001).
In terms of self-care, a majority of Type 1 and LADA respondents reported checking their blood
glucose five or more times per day; these respondents are far more likely than their peers with
other diabetes types to report doing so. Slight majorities reported seeing a diabetes nurse
educator in the past year and meeting with their diabetes physician in the past year. These
patterns did not differ by diabetes type. Fewer reported meeting with a nutritionist with no
difference across type.
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
7
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Table 2. Reported use of recommended diabetes preventive care practices, by diabetes type
N (%) OR for Type 1/LADA
(95% CI)
All
TuAnalyze
users N=931
Type 1
and
LADA
N=664
(71.3)
Type 2/Pre-
Diabetes
N=267 (28.7) Unadjusted
Adjusted for
age, sex,
race
Pneumonia shot ever
380 (40.8)
275
(41.4) 105 (39.3) 1.1 (.8, 1.5)
1.6 (1.1,
2.3)**
Flu shot/past year
606 (65.1) 459 (69) 148 (55.4)
1.8 (1.3,
2.4)***
2.3 (1.6,
3.3)***
Cholesterol
check/past year 837 (89.9)
595
(89.6) 242 (90.6) .9 (.5, 1.4) 1.7 (1, 3.1)
Eye exam/past year
766 (82.3)
580
(87.3) 186 (69.7)
3 (2.1,
4.3)***
4.8 (3.1,
7.5)***
Foot exam/past year
611 (65.6)
454
(68.4) 157 (58.8)
1.5 (1.1,
2)**
2.3 (1.6,
3.3)***
All 5 care practices
251 (30)
193
(29.1) 58 (21.7)
1.5 (1.1,
2.1)*
2.2 (1.5,
3.2)***
Self-monitors blood
glucose (SMBG) 5 or
more times per day 535 (57.5)
477
(71.8) 58 (21.7)
9.2 (6.6,
12.9)***
10.1 (6.8,
14.9)***
Nutrition visit/past
year 343 (36.8)
241
(36.3) 102 (38.2) .9 (.7, 1.2) .7 (.5, 1)
DNE visit/past year
525 (56.4)
383
(57.7) 142 (53.2) 1.3 (.9, 1.6) 1 (.7, 1.4)
4 or more MD
visits/past year 489 (52.5) 345 (52) 144 (53.9) .9 (.7. 1.2) 1 (.7, 1.4)
*p<.05 **p<.01 ***p<.001
Health status
A minority of users reported fair or poor health and approximately one-third report a most recent
A1c that is above the recommended target of 7%. One-quarter report a history of depression and
nearly half report any other comorbidity. No differences in health indicators across type were
found in adjusted analyses. Type 1 and LADA users were less likely to report a chronic
comorbid condition and more likely to have an above-target A1c in unadjusted analyses only.
Health status did not differ in relation to use of recommended preventive care measures in
analyses that adjusted for age, sex, race and type.
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
8
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Table 3. Health status, in aggregate and by type
N (%) OR for Type 1 (95% CI)
All users
(N=931)
Type 1 &
LADA
N=664
(71.3)
Type 2 & Pre-
Diabetes N=267
(28.7) Unadjusted
Adjusted for
age, sex, race
Poor self-rated
health 149 (16) 111 (16.7) 38 (14.2) 1.2 (.8, 1.8) 1.5 (.93, 2.3)
Depression 216 (23.2) 143 (21.5) 73 (27.3) .7 (.5, 1) .72 (.5, 1)
Any other
comorbidity 411 (44.2) 275 (41.4) 136 (50.9) .7 (.5, .9)** 1.1 (.78, 1.5)
A1c>7%
(N=826) 277 (33.5) 212 (35.8) 65 (27.8)
1.5 (1.04,
2)* 1.2 (.85, 1.8)
Comparison of TuAnalyze and BRFSS samples
In analyses of US-residing insulin-dependent respondents from both samples, TuAnalyze users
were more likely to be white and less likely to be over the age of 40 than the BRFSS sample,
confirming the extension of health monitoring into a different demographic. There was no
difference in the sex distribution across the two samples.
Table 4. Demographics of the TuAnalyze sample and the BRFSS subsample
TuAnalyze (N=577) BRFSSa (N=9,832) OR for TuA (95% CI)
White 510 (88.4) 6,693 (68.1) 3.6 (2.8, 4.6)***
Male 222 (38.5) 4,030 (41) .9 (.76, 1.1)
Over 40 325 (56.3) 9,245 (94) .08 (.07, .1)***
aNote that the BRFSS subsample comprises that portion of a national probability sample that
self-reports diabetes and insulin use.
In analyses that controlled for age, sex and race, TuAnalyze respondents were more likely than
BRFSS respondents to report they received an influenza immunization, eye exam, and all five
care practices in the past year. BRFSS respondents were more likely than the TuAnalyze
respondents to report they received an annual foot exam. The two samples differed greatly on
self-monitoring of blood glucose; nearly three quarters of the TuAnalyze sample and less than
one tenth of the national sample reported checking their blood sugar five or more times per day.
No difference was found in the history of pneumonia vaccination, annual lipid profile, or
frequency of doctor visits. Comparisons of diabetes educator or nutritionist visits were
precluded by the absence of these data in the BRFSS sample.
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
9
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Table 5. Comparison of reports of obtaining recommended diabetes preventive care
practices between insulin dependent TuAnalyze and BRFSS respondents
N (%) OR for TuAnalyze (95% CI)
TuAnalyze
(N=577)
BRFSS
(N=9,832)
Unadjusted
Adjusted for age,
sex, race
Pneumonia shot ever 291 (50.4) 6,262 (64.6) .56 (.47,
.66)***
1 (.84, 1.2)
Influenza immunization/
past year
415 (71.9) 6,537 (66.5) 1.3 (1.1, 1.6)* 1.9 (1.6, 2.3)***
Cholesterol check/past
year
525 (91) 8,769 (93.1) .75 (.56, 1) 1.1 (.78, 1.5)
Eye exam/past year 509 (88.2) 7,415 (75.4) 2.4 (1.9,
3.2)***
3.5 (2.6, 4.5)***
Foot exam/past year 420 (72.8) 7,777 (79.7) .68 (.56, .82)** .65 (.52, .79)***
All 5 care practices 208 (36.1) 3,264 (33.2) 1.1 (.95, 1.4) 1.8 (1.4, 2.1)***
SMBG 5 or more times
per day
422 (73.1) 903 (9.2) 26.9 (22.1,
32.8)***
13.3 (10.8,
16.5)***
Nutrition visit/past year 343 (36.8) -- -- --
DNE visit/past year 525 (56.4) -- -- --
4+ MD visits/past year 314 (54.4) 5,638 (57.3) .89 (.75, 1.05) .98 (.82, 1.2)
* p<.01 **p<.001 ***p<.0001
Discussion
Using a novel health monitoring approach, we collected information about preventive and self-
care practices from members of an international online diabetes social network. While a majority
of respondents appear to follow practice guidelines for specific preventive care services, less
than one third of the SN sample reported all five recommended practices—suggesting substantial
room for improvement in quality of care and disease management. Patterns vary by diabetes type
with higher levels of preventive care reported by respondents with Type 1 or LADA compared to
Type 2 or pre-diabetes. A similar pattern was seen for reports of appropriate SMBG. Results are
consistent with reports of type-based differences in service use, adherence and self-care
generated from studies of traditionally sampled cohorts (16-19).
Adherence to select healthcare practices reported by US-residing insulin dependent users of
TuAnalyze was mixed in relation to patterns found in a restricted comparator sample created
from the US national BRFSS. For the majority of comparison measures, SN application users
reported better use of preventive and self-care practices. This finding is not surprising given that
users of TuAnalyze and its host community TuDiabetes may include disproportionate
percentages of persons concerned with their health and with managing their diabetes—an artifact
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
10
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
of the self-selecting nature of the sample. It is notable however that even in this context there
were no differences across samples in levels of adherence to recommended quarterly doctor
visits, possibly reflecting insurance eligibility requirements in the US. Even allowing for the
healthy subject selection effect of the SN sample, there are sizeable gaps in care practice use
especially with regard to the composite measure of adherence to the five recommended care
practices.
Stepping back, this report adds to our previous finding that the SN mediated surveillance
approach can be used to engage distributed populations in health research by reporting about
current healthcare and self-care patterns by diabetes type. These measures are not available
through standard reporting systems. Learning how to monitor these issues using novel
approaches is, we contend, of high importance to public health given the large and growing
burden imposed by chronic illness. Effective monitoring of chronic as opposed to contagious
illness may benefit from tracking care patterns and management in detail and preservation of a
bidirectional communication channel with study samples for follow-up. Ensuring patient (or
sample) engagement may be especially important in this model.
Patient engagement in disease focused social networking is strong and growing (20). Growth
reflects the appeal of this organic and grassroots phenomenon, and patients’ need for community
(21), information and support (22-24). Harnessing this engagement for public health research
may be an important new direction for population health monitoring. Social networking is
increasingly common in the area of diabetes—about which there are many active online
communities of varying quality and safety (25). Opportunities are manifold to extend health
surveillance into these motivated and high value samples. The TuAnalyze approach
accomplishes this without sacrificing privacy, safety or the autonomy of individuals and their
communities.
Our approach is novel and findings should be read in the context of important limitations. Biases
in participation and validity of self-reported data are a focus of our research and they are intrinsic
to the model. Moreover, selection and participation biases are also present in more traditional
health monitoring systems. Comparisons of TuAnalyze and BRFSS data are novel however they
rely on assumptions about the adequacy of using self-reported insulin use as a proxy measure for
diabetes type and they reflect fundamentally different approaches to surveillance. We recognize
that these approaches are different and likely to yield different results. The social networking
medium is inherently open and dynamic and affords a bidirectional communication channel with
subjects. Application use and survey completion happen on a rolling basis that is in part indexed
to the overall growth of the community and changing uptake of the application. Given this, it is
challenging to ascertain a denominator that describes persons exposed to the site or active during
a given time period (4). This is acceptable when the goal is rapid and not representative health
surveillance—as befits this complementary monitoring mechanism.
Conclusion
Rapid assessment of preventive care practices and diabetes management strategies using a novel,
SN-mediated approach is feasible and can be used to fill gaps in traditional public health
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
11
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
monitoring of care practices by diabetes type. Our ability to harness this engagement without
sacrificing privacy or user control may provide an important new direction for public health
surveillance.
Acknowledgements
This work was supported by PO1HK000088-01 from the Centers for Disease Control and
Prevention (CDC).
Corresponding Author
Elissa R Weitzman, ScD, MSc
Children’s Hospital Informatics Program
One Autumn Street, Room 541
Boston, MA 02215 USA
Phone: +1 617 355 3538
Fax: +1 617 730 0267
Email: elissa.weitzman@childrens.harvard.edu
References
(1) Ramsey F, Ussery-Hall A, Garcia D, McDonald G, Easton A, Kambon M, Balluz L,
Garvin W, Vigeant J. Prevalence of selected risk behaviors and chronic diseases--
Behavioral Risk Factor Surveillance System (BRFSS), 39 steps communities, United
States, 2005. MMWR. Surveillance summaries : Morbidity and mortality weekly report.
Surveillance summaries / CDC. Oct 31 2008;57(11):1-20.
(2) Yach D, Hawkes C, Gould CL, Hofman KJ. The global burden of chronic diseases:
overcoming impediments to prevention and control. Jama. Jun 2 2004;291(21):2616-
2622.
(3) Bonander J, Gates S. Public health in an era of personal health records: opportunities for
innovation and new partnerships. Journal of medical Internet research. 2010;12(3):e33.
(4) Weitzman ER, Adida B, Kelemen S, Mandl KD. Sharing data for public health research
by members of an international online diabetes social network. PloS one.
2011;6(4):e19256.
(5) Hand E. Citizen science: People power. Nature. Aug 5;466(7307):685-687.
(6) Van Noort SP, Muehlen M, Rebelo de Andrade H, Koppeschaar C, Lima Lourenco JM,
Gomes MG. Gripenet: an internet-based system to monitor influenza-like illness
uniformly across Europe. Euro Surveill. Jul 2007;12(7):E5-6.
(7) Brownstein JS, Freifeld CC, Madoff LC. Digital disease detection--harnessing the Web
for public health surveillance. N Engl J Med. May 21 2009;360(21):2153-2155, 2157.
(8) A.D.A. Standards of Medical Care in Diabetes—2010. Diabetes Care.
2010;33(Supplement 1):S11-S61.
(9) Leslie RD, Kolb H, Schloot NC, Buzzetti R, Mauricio D, De Leiva A, Yderstraede K,
Sarti C, Thivolet C, Hadden D, Hunter S, Schernthaner G, Scherbaum W, Williams R,
Surveillance of an Online Social Network to Assess Population-level Diabetes Health Status and Healthcare Quality
12
Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * Vol.3, No. 3, 2011
Pozzilli P. Diabetes classification: grey zones, sound and smoke: Action LADA 1.
Diabetes/metabolism research and reviews. Oct 2008;24(7):511-519.
(10) Tuomi T. Type 1 and type 2 diabetes: what do they have in common? Diabetes. Dec
2005;54 Suppl 2:S40-45.
(11) Chowdhury P, Balluz L, Town M, Chowdhury FM, Bartolis W, Garvin W, Akcin H,
Greenlund KJ, Giles W. Surveillance of certain health behaviors and conditions among
states and selected local areas - Behavioral Risk Factor Surveillance System, United
States, 2007. MMWR. Surveillance summaries : Morbidity and mortality weekly report.
Surveillance summaries / CDC. Feb 5 2010;59(1):1-220.
(12) Saaddine JB, Engelgau MM, Beckles GL, Gregg EW, Thompson TJ, Narayan KM. A
diabetes report card for the United States: quality of care in the 1990s. Ann Intern Med.
Apr 16 2002;136(8):565-574.
(13) Cook C, Heath F, Thompson RL. A Meta-Analysis of Response Rates in Web- or
Internet-Based Surveys. Educational and Psychological Measurement. December 1, 2000
2000;60(6):821-836.
(14) Hoerger TJ, Segel JE, Gregg EW, Saaddine JB. Is glycemic control improving in U.S.
adults? Diabetes Care. Jan 2008;31(1):81-86.
(15) CDC. US Behavioral Risk Factor Surveillance System; 2009.
(16) Broadbent E, Donkin L, Stroh JC. Illness and treatment perceptions are associated with
adherence to medications, diet, and exercise in diabetic patients. Diabetes care. Feb
2011;34(2):338-340.
(17) Clarke WL. Behavioral challenges in the management of childhood diabetes. Journal of
diabetes science and technology. 2011;5(2):225-228.
(18) Kavookjian J, Elswick BM, Whetsel T. Interventions for being active among individuals
with diabetes: a systematic review of the literature. The Diabetes educator. Nov-Dec
2007;33(6):962-988; discussion 989-990.
(19) Peyrot M, Rubin RR, Kruger DF, Travis LB. Correlates of insulin injection omission.
Diabetes care. Feb 2010;33(2):240-245.
(20) Fox S. Peer‐to‐peer healthcare. Washington, D.C.: Pew Research Center; 2011.
(21) Frost JH, Massagli MP. Social uses of personal health information within
PatientsLikeMe, an online patient community: what can happen when patients have
access to one another's data. J Med Internet Res. 2008;10(3):e15.
(22) Seeman N. Web 2.0 and chronic illness: new horizons, new opportunities. Healthc Q.
2008;11(1):104-108, 110, 104.
(23) Chou WY, Hunt YM, Beckjord EB, Moser RP, Hesse BW. Social media use in the
United States: implications for health communication. J Med Internet Res.
2009;11(4):e48.
(24) Idriss SZ, Kvedar JC, Watson AJ. The role of online support communities: benefits of
expanded social networks to patients with psoriasis. Arch Dermatol. Jan 2009;145(1):46-
51.
(25) Weitzman ER, Cole E, Kaci L, Mandl KD. Social but safe? Quality and safety of
diabetes-related online social networks. J Am Med Inform Assoc. Jan 24 2011.